out1 / checkpoint-135 /trainer_state.json
hardlyworking's picture
Upload folder using huggingface_hub
4ff0dd8 verified
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9967697277341947,
"eval_steps": 500,
"global_step": 135,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.007383479464697739,
"grad_norm": 1.9591929912567139,
"learning_rate": 0.0,
"loss": 1.6228,
"memory/device_mem_reserved(gib)": 21.61,
"memory/max_mem_active(gib)": 21.2,
"memory/max_mem_allocated(gib)": 21.2,
"step": 1
},
{
"epoch": 0.014766958929395477,
"grad_norm": 1.4523507356643677,
"learning_rate": 1.5384615384615387e-05,
"loss": 1.5769,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 2
},
{
"epoch": 0.022150438394093218,
"grad_norm": 1.1918187141418457,
"learning_rate": 3.0769230769230774e-05,
"loss": 1.5435,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 3
},
{
"epoch": 0.029533917858790955,
"grad_norm": 0.8260876536369324,
"learning_rate": 4.615384615384616e-05,
"loss": 1.6523,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 4
},
{
"epoch": 0.03691739732348869,
"grad_norm": 0.8584926128387451,
"learning_rate": 6.153846153846155e-05,
"loss": 1.5745,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 5
},
{
"epoch": 0.044300876788186436,
"grad_norm": 0.6466429829597473,
"learning_rate": 7.692307692307693e-05,
"loss": 1.4759,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 6
},
{
"epoch": 0.05168435625288417,
"grad_norm": 0.5014482140541077,
"learning_rate": 9.230769230769232e-05,
"loss": 1.602,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 7
},
{
"epoch": 0.05906783571758191,
"grad_norm": 0.6017433404922485,
"learning_rate": 0.0001076923076923077,
"loss": 1.4176,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 8
},
{
"epoch": 0.06645131518227965,
"grad_norm": 0.4612258970737457,
"learning_rate": 0.0001230769230769231,
"loss": 1.5819,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 9
},
{
"epoch": 0.07383479464697738,
"grad_norm": 0.4430214464664459,
"learning_rate": 0.00013846153846153847,
"loss": 1.561,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 10
},
{
"epoch": 0.08121827411167512,
"grad_norm": 0.3746771216392517,
"learning_rate": 0.00015384615384615385,
"loss": 1.6744,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 11
},
{
"epoch": 0.08860175357637287,
"grad_norm": 0.38248857855796814,
"learning_rate": 0.00016923076923076923,
"loss": 1.5629,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 12
},
{
"epoch": 0.09598523304107061,
"grad_norm": 0.515844464302063,
"learning_rate": 0.00018461538461538463,
"loss": 1.5264,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 13
},
{
"epoch": 0.10336871250576835,
"grad_norm": 0.3964424431324005,
"learning_rate": 0.0002,
"loss": 1.5398,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 14
},
{
"epoch": 0.11075219197046608,
"grad_norm": 0.4010593891143799,
"learning_rate": 0.0001999668467514313,
"loss": 1.4618,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 15
},
{
"epoch": 0.11813567143516382,
"grad_norm": 0.3192802965641022,
"learning_rate": 0.00019986740898848306,
"loss": 1.6994,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 16
},
{
"epoch": 0.12551915089986157,
"grad_norm": 0.410099059343338,
"learning_rate": 0.00019970175264485266,
"loss": 1.5913,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 17
},
{
"epoch": 0.1329026303645593,
"grad_norm": 0.312429815530777,
"learning_rate": 0.0001994699875614589,
"loss": 1.5701,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 18
},
{
"epoch": 0.14028610982925704,
"grad_norm": 0.2831230163574219,
"learning_rate": 0.00019917226741361015,
"loss": 1.5744,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 19
},
{
"epoch": 0.14766958929395477,
"grad_norm": 0.3618868291378021,
"learning_rate": 0.00019880878960910772,
"loss": 1.5185,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 20
},
{
"epoch": 0.15505306875865252,
"grad_norm": 0.3151628077030182,
"learning_rate": 0.00019837979515735166,
"loss": 1.5086,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 21
},
{
"epoch": 0.16243654822335024,
"grad_norm": 0.31955838203430176,
"learning_rate": 0.0001978855685095358,
"loss": 1.6329,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 22
},
{
"epoch": 0.169820027688048,
"grad_norm": 0.3030437231063843,
"learning_rate": 0.00019732643737003827,
"loss": 1.6697,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 23
},
{
"epoch": 0.17720350715274574,
"grad_norm": 0.41288134455680847,
"learning_rate": 0.00019670277247913205,
"loss": 1.7094,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 24
},
{
"epoch": 0.18458698661744347,
"grad_norm": 0.2887294888496399,
"learning_rate": 0.00019601498736716017,
"loss": 1.5554,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 25
},
{
"epoch": 0.19197046608214122,
"grad_norm": 0.3173791170120239,
"learning_rate": 0.00019526353808033825,
"loss": 1.4404,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 26
},
{
"epoch": 0.19935394554683894,
"grad_norm": 0.2877439558506012,
"learning_rate": 0.00019444892287836613,
"loss": 1.4766,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 27
},
{
"epoch": 0.2067374250115367,
"grad_norm": 0.29286038875579834,
"learning_rate": 0.00019357168190404936,
"loss": 1.5156,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 28
},
{
"epoch": 0.2141209044762344,
"grad_norm": 0.27713659405708313,
"learning_rate": 0.00019263239682514952,
"loss": 1.5153,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 29
},
{
"epoch": 0.22150438394093216,
"grad_norm": 0.29187655448913574,
"learning_rate": 0.0001916316904487005,
"loss": 1.6036,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 30
},
{
"epoch": 0.22888786340562992,
"grad_norm": 0.2671583890914917,
"learning_rate": 0.00019057022630804716,
"loss": 1.4675,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 31
},
{
"epoch": 0.23627134287032764,
"grad_norm": 0.2679831087589264,
"learning_rate": 0.00018944870822287956,
"loss": 1.581,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 32
},
{
"epoch": 0.2436548223350254,
"grad_norm": 0.26359617710113525,
"learning_rate": 0.00018826787983255473,
"loss": 1.4674,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 33
},
{
"epoch": 0.25103830179972314,
"grad_norm": 0.30446046590805054,
"learning_rate": 0.00018702852410301554,
"loss": 1.5038,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 34
},
{
"epoch": 0.25842178126442084,
"grad_norm": 0.3004315197467804,
"learning_rate": 0.00018573146280763324,
"loss": 1.4024,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 35
},
{
"epoch": 0.2658052607291186,
"grad_norm": 0.27353399991989136,
"learning_rate": 0.00018437755598231856,
"loss": 1.3652,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 36
},
{
"epoch": 0.27318874019381634,
"grad_norm": 0.2659265995025635,
"learning_rate": 0.0001829677013552619,
"loss": 1.4905,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 37
},
{
"epoch": 0.2805722196585141,
"grad_norm": 0.2703750431537628,
"learning_rate": 0.00018150283375168114,
"loss": 1.3298,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 38
},
{
"epoch": 0.28795569912321184,
"grad_norm": 0.29322877526283264,
"learning_rate": 0.00017998392447397197,
"loss": 1.3639,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 39
},
{
"epoch": 0.29533917858790953,
"grad_norm": 0.26927265524864197,
"learning_rate": 0.00017841198065767107,
"loss": 1.4849,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 40
},
{
"epoch": 0.3027226580526073,
"grad_norm": 0.26683205366134644,
"learning_rate": 0.00017678804460366,
"loss": 1.4523,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 41
},
{
"epoch": 0.31010613751730504,
"grad_norm": 0.26909339427948,
"learning_rate": 0.00017511319308705198,
"loss": 1.6047,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 42
},
{
"epoch": 0.3174896169820028,
"grad_norm": 0.2938016355037689,
"learning_rate": 0.00017338853664321992,
"loss": 1.3858,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 43
},
{
"epoch": 0.3248730964467005,
"grad_norm": 0.28290048241615295,
"learning_rate": 0.00017161521883143934,
"loss": 1.3671,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 44
},
{
"epoch": 0.33225657591139823,
"grad_norm": 0.2870519161224365,
"learning_rate": 0.00016979441547663435,
"loss": 1.5334,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 45
},
{
"epoch": 0.339640055376096,
"grad_norm": 0.26375913619995117,
"learning_rate": 0.00016792733388972932,
"loss": 1.5627,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 46
},
{
"epoch": 0.34702353484079373,
"grad_norm": 0.26975446939468384,
"learning_rate": 0.00016601521206712318,
"loss": 1.4987,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 47
},
{
"epoch": 0.3544070143054915,
"grad_norm": 0.30166226625442505,
"learning_rate": 0.00016405931786981755,
"loss": 1.5128,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 48
},
{
"epoch": 0.3617904937701892,
"grad_norm": 0.29190436005592346,
"learning_rate": 0.00016206094818274229,
"loss": 1.4914,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 49
},
{
"epoch": 0.36917397323488693,
"grad_norm": 0.2847207486629486,
"learning_rate": 0.00016002142805483685,
"loss": 1.4755,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 50
},
{
"epoch": 0.3765574526995847,
"grad_norm": 0.2877140939235687,
"learning_rate": 0.00015794210982045636,
"loss": 1.422,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 51
},
{
"epoch": 0.38394093216428243,
"grad_norm": 0.2705255150794983,
"learning_rate": 0.00015582437220268647,
"loss": 1.4822,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 52
},
{
"epoch": 0.3913244116289802,
"grad_norm": 0.2597866654396057,
"learning_rate": 0.00015366961939916008,
"loss": 1.5088,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 53
},
{
"epoch": 0.3987078910936779,
"grad_norm": 0.2965547740459442,
"learning_rate": 0.0001514792801509831,
"loss": 1.5346,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 54
},
{
"epoch": 0.40609137055837563,
"grad_norm": 0.2670862078666687,
"learning_rate": 0.00014925480679538647,
"loss": 1.5695,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 55
},
{
"epoch": 0.4134748500230734,
"grad_norm": 0.30679062008857727,
"learning_rate": 0.000146997674302732,
"loss": 1.6911,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 56
},
{
"epoch": 0.42085832948777113,
"grad_norm": 0.2812528908252716,
"learning_rate": 0.0001447093792985114,
"loss": 1.664,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 57
},
{
"epoch": 0.4282418089524688,
"grad_norm": 0.26079821586608887,
"learning_rate": 0.0001423914390709861,
"loss": 1.3911,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 58
},
{
"epoch": 0.4356252884171666,
"grad_norm": 0.3026648461818695,
"learning_rate": 0.00014004539056512667,
"loss": 1.5685,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 59
},
{
"epoch": 0.44300876788186433,
"grad_norm": 0.28435277938842773,
"learning_rate": 0.00013767278936351854,
"loss": 1.3599,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 60
},
{
"epoch": 0.4503922473465621,
"grad_norm": 0.29877325892448425,
"learning_rate": 0.0001352752086549095,
"loss": 1.5173,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 61
},
{
"epoch": 0.45777572681125983,
"grad_norm": 0.29666033387184143,
"learning_rate": 0.0001328542381910835,
"loss": 1.5017,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 62
},
{
"epoch": 0.4651592062759575,
"grad_norm": 0.26934438943862915,
"learning_rate": 0.0001304114832327518,
"loss": 1.4674,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 63
},
{
"epoch": 0.4725426857406553,
"grad_norm": 0.30038923025131226,
"learning_rate": 0.00012794856348516095,
"loss": 1.5306,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 64
},
{
"epoch": 0.47992616520535303,
"grad_norm": 0.2799171805381775,
"learning_rate": 0.00012546711202412287,
"loss": 1.4696,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 65
},
{
"epoch": 0.4873096446700508,
"grad_norm": 0.313919335603714,
"learning_rate": 0.0001229687742131796,
"loss": 1.5927,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 66
},
{
"epoch": 0.4946931241347485,
"grad_norm": 0.27579808235168457,
"learning_rate": 0.0001204552066126201,
"loss": 1.4654,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 67
},
{
"epoch": 0.5020766035994463,
"grad_norm": 0.28434908390045166,
"learning_rate": 0.00011792807588107357,
"loss": 1.5039,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 68
},
{
"epoch": 0.509460083064144,
"grad_norm": 0.2828267514705658,
"learning_rate": 0.0001153890576704062,
"loss": 1.6203,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 69
},
{
"epoch": 0.5168435625288417,
"grad_norm": 0.30293530225753784,
"learning_rate": 0.00011283983551465511,
"loss": 1.5628,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 70
},
{
"epoch": 0.5242270419935394,
"grad_norm": 0.2633240520954132,
"learning_rate": 0.00011028209971373605,
"loss": 1.5666,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 71
},
{
"epoch": 0.5316105214582372,
"grad_norm": 0.24525777995586395,
"learning_rate": 0.00010771754621266466,
"loss": 1.592,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 72
},
{
"epoch": 0.5389940009229349,
"grad_norm": 0.27619901299476624,
"learning_rate": 0.00010514787547703466,
"loss": 1.6212,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 73
},
{
"epoch": 0.5463774803876327,
"grad_norm": 0.2783447504043579,
"learning_rate": 0.00010257479136549889,
"loss": 1.457,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 74
},
{
"epoch": 0.5537609598523304,
"grad_norm": 0.2755623459815979,
"learning_rate": 0.0001,
"loss": 1.3652,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 75
},
{
"epoch": 0.5611444393170282,
"grad_norm": 0.28049108386039734,
"learning_rate": 9.742520863450115e-05,
"loss": 1.4187,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 76
},
{
"epoch": 0.5685279187817259,
"grad_norm": 0.27499496936798096,
"learning_rate": 9.485212452296535e-05,
"loss": 1.5488,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 77
},
{
"epoch": 0.5759113982464237,
"grad_norm": 0.288296639919281,
"learning_rate": 9.228245378733537e-05,
"loss": 1.412,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 78
},
{
"epoch": 0.5832948777111213,
"grad_norm": 0.35849323868751526,
"learning_rate": 8.971790028626395e-05,
"loss": 1.5214,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 79
},
{
"epoch": 0.5906783571758191,
"grad_norm": 0.2951800525188446,
"learning_rate": 8.71601644853449e-05,
"loss": 1.4837,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 80
},
{
"epoch": 0.5980618366405168,
"grad_norm": 0.2763918340206146,
"learning_rate": 8.461094232959381e-05,
"loss": 1.5419,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 81
},
{
"epoch": 0.6054453161052146,
"grad_norm": 0.2745305895805359,
"learning_rate": 8.207192411892646e-05,
"loss": 1.4921,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 82
},
{
"epoch": 0.6128287955699123,
"grad_norm": 0.30081912875175476,
"learning_rate": 7.954479338737995e-05,
"loss": 1.4745,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 83
},
{
"epoch": 0.6202122750346101,
"grad_norm": 0.27640482783317566,
"learning_rate": 7.703122578682046e-05,
"loss": 1.4461,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 84
},
{
"epoch": 0.6275957544993078,
"grad_norm": 0.3407949209213257,
"learning_rate": 7.453288797587714e-05,
"loss": 1.617,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 85
},
{
"epoch": 0.6349792339640056,
"grad_norm": 0.2856479287147522,
"learning_rate": 7.205143651483906e-05,
"loss": 1.6538,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 86
},
{
"epoch": 0.6423627134287033,
"grad_norm": 0.39668434858322144,
"learning_rate": 6.958851676724823e-05,
"loss": 1.5753,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 87
},
{
"epoch": 0.649746192893401,
"grad_norm": 0.28521108627319336,
"learning_rate": 6.714576180891654e-05,
"loss": 1.3448,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 88
},
{
"epoch": 0.6571296723580987,
"grad_norm": 0.2806543707847595,
"learning_rate": 6.472479134509052e-05,
"loss": 1.6431,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 89
},
{
"epoch": 0.6645131518227965,
"grad_norm": 0.28897204995155334,
"learning_rate": 6.232721063648148e-05,
"loss": 1.5744,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 90
},
{
"epoch": 0.6718966312874942,
"grad_norm": 0.25295817852020264,
"learning_rate": 5.9954609434873344e-05,
"loss": 1.4552,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 91
},
{
"epoch": 0.679280110752192,
"grad_norm": 0.2824646234512329,
"learning_rate": 5.7608560929013946e-05,
"loss": 1.4193,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 92
},
{
"epoch": 0.6866635902168897,
"grad_norm": 0.27227288484573364,
"learning_rate": 5.5290620701488594e-05,
"loss": 1.4602,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 93
},
{
"epoch": 0.6940470696815875,
"grad_norm": 0.26284658908843994,
"learning_rate": 5.300232569726804e-05,
"loss": 1.5947,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 94
},
{
"epoch": 0.7014305491462852,
"grad_norm": 0.26174792647361755,
"learning_rate": 5.074519320461357e-05,
"loss": 1.4122,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 95
},
{
"epoch": 0.708814028610983,
"grad_norm": 0.3069747984409332,
"learning_rate": 4.852071984901696e-05,
"loss": 1.4805,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 96
},
{
"epoch": 0.7161975080756806,
"grad_norm": 0.32007896900177,
"learning_rate": 4.633038060083996e-05,
"loss": 1.3798,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 97
},
{
"epoch": 0.7235809875403784,
"grad_norm": 0.26006585359573364,
"learning_rate": 4.417562779731355e-05,
"loss": 1.4974,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 98
},
{
"epoch": 0.7309644670050761,
"grad_norm": 0.2552710473537445,
"learning_rate": 4.205789017954364e-05,
"loss": 1.4851,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 99
},
{
"epoch": 0.7383479464697739,
"grad_norm": 0.23933957517147064,
"learning_rate": 3.997857194516319e-05,
"loss": 1.5046,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 100
},
{
"epoch": 0.7457314259344716,
"grad_norm": 0.2627595067024231,
"learning_rate": 3.793905181725772e-05,
"loss": 1.4958,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 101
},
{
"epoch": 0.7531149053991694,
"grad_norm": 0.3138872981071472,
"learning_rate": 3.594068213018249e-05,
"loss": 1.4714,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 102
},
{
"epoch": 0.7604983848638671,
"grad_norm": 0.2678472697734833,
"learning_rate": 3.3984787932876814e-05,
"loss": 1.5767,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 103
},
{
"epoch": 0.7678818643285649,
"grad_norm": 0.29931122064590454,
"learning_rate": 3.207266611027069e-05,
"loss": 1.5006,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 104
},
{
"epoch": 0.7752653437932626,
"grad_norm": 0.27635884284973145,
"learning_rate": 3.0205584523365626e-05,
"loss": 1.4236,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 105
},
{
"epoch": 0.7826488232579604,
"grad_norm": 0.2608022093772888,
"learning_rate": 2.8384781168560693e-05,
"loss": 1.3451,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 106
},
{
"epoch": 0.790032302722658,
"grad_norm": 0.271505743265152,
"learning_rate": 2.6611463356780096e-05,
"loss": 1.5161,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 107
},
{
"epoch": 0.7974157821873558,
"grad_norm": 0.26420703530311584,
"learning_rate": 2.4886806912948035e-05,
"loss": 1.5567,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 108
},
{
"epoch": 0.8047992616520535,
"grad_norm": 0.2610979974269867,
"learning_rate": 2.3211955396340002e-05,
"loss": 1.6357,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 109
},
{
"epoch": 0.8121827411167513,
"grad_norm": 0.25315845012664795,
"learning_rate": 2.1588019342328968e-05,
"loss": 1.4353,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 110
},
{
"epoch": 0.819566220581449,
"grad_norm": 0.27436307072639465,
"learning_rate": 2.0016075526028065e-05,
"loss": 1.4613,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 111
},
{
"epoch": 0.8269497000461468,
"grad_norm": 0.2581930458545685,
"learning_rate": 1.8497166248318876e-05,
"loss": 1.4317,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 112
},
{
"epoch": 0.8343331795108445,
"grad_norm": 0.3470868170261383,
"learning_rate": 1.703229864473811e-05,
"loss": 1.5637,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 113
},
{
"epoch": 0.8417166589755423,
"grad_norm": 0.2554042339324951,
"learning_rate": 1.562244401768144e-05,
"loss": 1.597,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 114
},
{
"epoch": 0.84910013844024,
"grad_norm": 0.26158565282821655,
"learning_rate": 1.426853719236676e-05,
"loss": 1.3765,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 115
},
{
"epoch": 0.8564836179049377,
"grad_norm": 0.27990010380744934,
"learning_rate": 1.2971475896984475e-05,
"loss": 1.2698,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 116
},
{
"epoch": 0.8638670973696354,
"grad_norm": 0.2723514139652252,
"learning_rate": 1.1732120167445248e-05,
"loss": 1.5009,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 117
},
{
"epoch": 0.8712505768343332,
"grad_norm": 0.2686865031719208,
"learning_rate": 1.0551291777120464e-05,
"loss": 1.4692,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 118
},
{
"epoch": 0.8786340562990309,
"grad_norm": 0.2559969127178192,
"learning_rate": 9.429773691952858e-06,
"loss": 1.4417,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 119
},
{
"epoch": 0.8860175357637287,
"grad_norm": 0.3991638123989105,
"learning_rate": 8.368309551299536e-06,
"loss": 1.5384,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 120
},
{
"epoch": 0.8934010152284264,
"grad_norm": 0.2762066125869751,
"learning_rate": 7.367603174850502e-06,
"loss": 1.4997,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 121
},
{
"epoch": 0.9007844946931242,
"grad_norm": 0.2837660014629364,
"learning_rate": 6.428318095950647e-06,
"loss": 1.448,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 122
},
{
"epoch": 0.9081679741578219,
"grad_norm": 0.2550673484802246,
"learning_rate": 5.551077121633874e-06,
"loss": 1.4488,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 123
},
{
"epoch": 0.9155514536225197,
"grad_norm": 0.2584616541862488,
"learning_rate": 4.7364619196617495e-06,
"loss": 1.3641,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 124
},
{
"epoch": 0.9229349330872173,
"grad_norm": 0.2741107940673828,
"learning_rate": 3.985012632839824e-06,
"loss": 1.4923,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 125
},
{
"epoch": 0.930318412551915,
"grad_norm": 0.252888023853302,
"learning_rate": 3.2972275208679625e-06,
"loss": 1.4971,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 126
},
{
"epoch": 0.9377018920166128,
"grad_norm": 0.25903505086898804,
"learning_rate": 2.6735626299617457e-06,
"loss": 1.3392,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 127
},
{
"epoch": 0.9450853714813106,
"grad_norm": 0.2537846267223358,
"learning_rate": 2.1144314904642195e-06,
"loss": 1.3883,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 128
},
{
"epoch": 0.9524688509460083,
"grad_norm": 0.2762198746204376,
"learning_rate": 1.6202048426483651e-06,
"loss": 1.4447,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 129
},
{
"epoch": 0.9598523304107061,
"grad_norm": 0.25517532229423523,
"learning_rate": 1.1912103908922945e-06,
"loss": 1.455,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 130
},
{
"epoch": 0.9672358098754038,
"grad_norm": 0.2435961812734604,
"learning_rate": 8.277325863898511e-07,
"loss": 1.4236,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 131
},
{
"epoch": 0.9746192893401016,
"grad_norm": 0.28410804271698,
"learning_rate": 5.300124385410943e-07,
"loss": 1.4976,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 132
},
{
"epoch": 0.9820027688047993,
"grad_norm": 0.2661467492580414,
"learning_rate": 2.9824735514732974e-07,
"loss": 1.3701,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 133
},
{
"epoch": 0.989386248269497,
"grad_norm": 0.2969636023044586,
"learning_rate": 1.3259101151694708e-07,
"loss": 1.6409,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 134
},
{
"epoch": 0.9967697277341947,
"grad_norm": 0.26862043142318726,
"learning_rate": 3.3153248568695835e-08,
"loss": 1.5537,
"memory/device_mem_reserved(gib)": 21.8,
"memory/max_mem_active(gib)": 21.55,
"memory/max_mem_allocated(gib)": 21.55,
"step": 135
}
],
"logging_steps": 1,
"max_steps": 135,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 34,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 2.4510152212217856e+18,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}